This invention relates to a crucible for growing large macrocrystals (or ingots) from a melt in a highly modified Stockbarger type furnace, or a comparable one in which slow growth of a solidifying macrocrystal from the melt under the influence of the earth""s gravity, results in the solidified macrocrystal being supported within the walls of the crucible. It is accepted practice to refer to melt-grown isometric blanks as being xe2x80x9csinglexe2x80x9d when the finished crystal does not exhibit flaws attributable to multiple components disregarding mosaics spread which is never less than 0.25xc2x0 in the flattest cleavage and can generally be seen in the texture of thermal etch. In fact a mosaic spread of 3xc2x0 is valued in pulled crystals for not showing the following flaws attributable to multiple components though their displacement may be less than 1xc2x0. However the displacement may be any angle up to 90xc2x0 and still be optically homogeneous when not decorated by inclusions.
Multiple components are said to be present when they are visible as a fringe pattern, or strain pattern in an optically polished crystal. Multiplicities will fracture non-cubic crystals while mosaics will not. Mosaic spread is important for X-ray plates and gun sights. Mosaics which do not have unacceptable flaws are acceptable in a single crystal.
An ideal macrocrystal is a single crystal, but typically a macrocrystal grown is only optically monocrystalline, being formed of several crystals demarcated by boundaries within the macrocrystal. Grain boundaries within a large macrocrystal (an ingot) which are discontinuities are undesirable.
More specifically the novel crucible is useful to grow crystals of halides of an element of Group 1a and Group 2a of the Periodic Table, particularly the alkali metal fluorides and alkaline earth metal fluorides, lead fluoride and crystals of the foregoing salts which are doped with desirable metal ion dopands. High quality macrocrystals of fluorides of lithium and sodium, as well as of magnesium, calcium, barium and strontium have been produced commercially since soon after 1939 when Donald Stockbarger disclosed a method for their manufacture in U.S. Pat. No. 2,149,076 and we taught methods for purifying melts in U.S. Pat. Nos. 2,498,186 and 2,550,173. However, producing high quality macrocrystals depends upon many factors, including having the skills required to duplicate successful runs. Success depends upon anticipating the need for minute adjustments in xe2x80x9cpower hoursxe2x80x9d (rate at which power is delivered to heat the melt), among others, all of which adjustments must be made before a probe inserted in the melt shows any indication of a change in the rate of growth.
To date, a graphite crucible may be used for growing a Stockbarger macro-crystal, provided the graphite is not so porous as to allow the melt to leak through it. A crucible may have an inclined or flat bottom and either might be provided with a well for a xe2x80x9cseed crystalxe2x80x9d holder. If a seed is used, it is incorporated into the bottom of the macrocrystal formed.
U.S. Pat. No. 5,911,824 teaches that a particular graphite crucible could not contain a melt of thallium iodide doped sodium iodide, NaI(TI) (see col 4, lines 26-34). A NaI(Tl) crystal was successfully grown in a graphite crucible the inside surface of which is coated with graphitic pyrolitic carbon, and the crystal did not adhere to the coated surface (see col 4, lines 36-47).
However, a second crystal grown in the same crucible, adhered strongly to the crucible and could not be removed in its entirety without a remelting procedure (see col 4, lines 60-66). If the crystal was grown in a platinum crucible, the crystal adheres to the platinum. The platinum crucible must be heated to melt the surfaces of the crystal in contact with the platinum (referred to as xe2x80x9cremeltingxe2x80x9d) before the platinum crucible may be slid off the crystal.
It is a characteristic of macrocrystals of the aforementioned fluorides of melt-grown metal or metal-like elements, that they shrink when they solidify (referred to herein as xe2x80x9cshrinking meltsxe2x80x9d), and the first crystal grown may be readily removed because the walls of a graphite or carbon crucible are not wetted by the melt.
To be readily removed, sufficient shrinkage must occur when, and after, the melt solidifies provided also that the bond between crystal and crucible is weaker than the force needed to fracture the crystal as it cools. Since the coefficient of linear expansion of graphite is about 7.85 microinches/(inch)(xc2x0 C.) measured at about 40xc2x0 C., that of the salt must be greater, preferably at least 5% greater, and more preferably 10% greater. The coefficient of fluorspar is about 19.5 microinches/(inch)(xc2x0 C.) measured at about 40xc2x0 C., so that even in a relatively small crucible having an inside diameter in the range from about 5 cm to about 25 cm, the circumferential surface of the solidified crystal pulls away from the inner walls of the crucible, and the crystal may be lifted out of the crucible with a vacuum cup without breaking the crystal or crucible, even when the crucible is cylindrical and its sides are vertical. This may be done provided the surface tension of the melt is high enough so as to fail to substantially wet the walls of the crucible; because the melt does not wet the surface of the graphite it does not to seep into its pores. Note that the linear coefficients of expansion at melt temperature will likely be substantially different from those given above.
The effect of such shrinkage on a crystal no larger than about 7.5 cm at its greatest diameter (referred to as a 7.5 cm diameter macrocrystal) is not particularly notable even if the shrinkage is not controlled, but for larger crystals that effect is; the larger the crystal, the more detrimental are the effects of such uncontrolled shrinkage.
Using a shrinking melt to grow a Stockbarger macrocrystal typically comprises slowly moving the melt at a controlled rate from a region hotter than its solidification temperature to a region cooler than its solidification temperature, controlling the relative temperatures of the regions, and maintaining a temperature gradient in a localized zone between the regions at the boundary of the melt. The temperature gradient in the zone is sufficient to allow melt to crystallize at the cooler boundary of the localized zone. An xe2x80x9celevatorxe2x80x9d type furnace may be used where the crucible is raised or lowered on an elevator; or a xe2x80x9cmovable temperature gradientxe2x80x9d furnace may be used where the furnace is moved and the crucible is stationary. The gradient between melt near the top of a crystal and the sharply localized zone is in the range from about 100xc2x0 C. to about 500xc2x0 C. depending upon the particular halide.
It will be evident that the temperature of the edge portion of successive layers of the melt corresponds to the solidification temperature of the melt as these edge portions reach a substantially fixed location in the path of travel of the melt and solidification begins and progresses inwardly. Preferred crystals are obtained when the zone of solidification approximates a plane. It is desirable to control the rate of heat flow through the inner portion of the melt from the hotter to the cooler region. If the rate of heat flow through the inner portion of the melt is too slow, the zone of solidification tends to be concave. Properly controlling the rate of heat flow through the inner portion of the melt allows the zone of solidification to approximate a plane.
To date, the art has addressed the problem of a macrocrystal adhering to the inner surface of the crucible by either providing a very smooth microporous graphite surface, or by coating the surface of the graphite as in the ""824 patent, or by lining a mechanically stable temperature resistant material such as alundum or graphite with a thin sheet of platinum as in U.S. Pat. No. 5,997,640. In either case, the better is the separation upon solidification, the more readily the macrocrystal falls to the bottom of the crucible. In a crucible in which the porosity is such that the melt seeps partially into the pores for a short distance sufficient to hold the weight of the crystal, the bottom of the solidified crystal pulls away from the bottom of the crucible, leaving a bottom gap. In many instances the seepage is insufficient to support the weight of the crystal in the walls of the crucible and the crystal will suddenly drop to the bottom of the crucible closing the gap. Such partial seepage into the pores is deemed to fail substantially to wet the interior surface of the crucible.
Making an uncontrolled change in the flow of heat and the rate of growth of the crystal within the crucible gives rise to xe2x80x9csparklesxe2x80x9d. Sparkles are especially evident in ingots larger than 7.5 cm effective diameter, this being the longest diameter in a horizontal plane. With effective diameters smaller than about 7.5 cm, either the ingot does not drop or the flaws are lost with others in the xe2x80x9cheelxe2x80x9d of the ingot. Small ingots generally lack sparkles because minute, micron-size gas inclusions must migrate upwards to gain in size sufficient to develop reflecting facets and become visible. The xe2x80x9cheelxe2x80x9d refers to the base of an ingot which has been removed from the crucible in which the ingot was grown, and placed upside-down, so that the ingot rests on its upper surface; while the ingot was being formed the heel was at the top of the ingot. The upper portion refers to a zone above the mid-horizontal plane through the macrocrystal; the upper portion terminates in a heel demarcatable from the upper portion. The upper portion of an ingot grown in the novel crucible is essentially free from gas-induced sparkles.
In many instances the number of multiplicities is not of great concern, but in those instances where an essentially single crystal is desired, it is generally hoped that number is minimized by seeding with a single crystal. The novel crucible of this invention provides a far more reliable means for minimizing the number of multiplicities in an ingot, particularly relatively large ones which to date, are never single.
Relatively large ingots, particularly those larger than about 15 cm (6 ins.) in effective diameter, are highly susceptible to flaws due to gas inclusions, caused by small pockets of gases trapped within the ingot as it is solidifying. Such inclusions are referred to in the art as xe2x80x9csparklesxe2x80x9d because these inclusions reflect flashes of light, or glitter, within the crystal when light is shone through the crystal. It is well established that sparkles are voids bounded by flat 111 facets because polarized light shows no strain that even the smallest solid would generate from different contraction rates. In CaF2 grown in a Stockbarger furnace, sparkles are evidenced by flat six-sided facets. Migration of the voids (upward in a Stockbarger ingot) toward the hotter zone allows these voids to combine and spread, forming a broad generally horizontal band. In relatively small barium fluoride (BaF2) ingots 15 cm (6 ins) in diameter some sparkles grow to 5 mm in width while others remain less than 1 xcexcm wide, all reflecting from 111 facets set by the host crystal.
In an essentially perfect crystal lattice there is no tolerable physical location in the lattice of ordered atoms for gas molecules. Any interruption which causes faster growth will trap dissolved gases on a molecular scale. While the crystal is cooling, and within about 50xc2x0 C. of its melting point, trapped molecules of gases migrate toward the hotter zone. Such molecules of trapped gases are mainly carbon monoxide (CO) and hydrogen (H2). Growing an ingot of an alkali metal or alkaline earth metal fluoride in a graphite crucible may provide, in addition, carbon tetrafluoride (CF4) and other fluorocarbons probably by the action of PbF2 (used as a getter) vapor on the very hot graphite heater along with molecules of nitrogen (N2) from the air. Leakage of air is held to a minimum to avoid oxygen-stabilized color centers in ingots of alkaline earth fluorides, and others. Carbon monoxide (CO) and CF4 should be more soluble in these melts than hydrogen (H2).
In a platinum crucible, the ingot shrinks after it solidifies but still adheres to the inner surface of the platinum crucible, contracting not only the sides but also the bottom of the crucible. This is the reason why the platinum crucible is heated sufficiently to free the ingot so that its contraction will not crack it. Just the contraction during growth shrinks the platinum crucible so that it must be rolled on a mandrel to stretch it to its initial size before it is reused. I concluded that the more uniform and better heat transfer through the walls and bottom of the platinum crucible was the reason why an ingot grown in a platinum crucible was substantially free from xe2x80x9csparklesxe2x80x9d caused by gas inclusions. Also, adhesion of the solidifying ingot to the platinum prevents any movement inside the ingot during its growth.
It should be noted that an ingot grown in a platinum crucible may have other, non-gas derived inclusions, also referred to as xe2x80x9csparklesxe2x80x9d, generated by the intense radiation of the xe2x80x9cremelt furnacexe2x80x9d (used specifically to remelt the adhering surface of the ingot) when, after the platinum crucible is removed, the ingot continues to be heated. Such sparkles are the result of the recrystallization of a small volume of melt generated within the solidified ingot by radiant heat. Such xe2x80x9cnon-gas derivedxe2x80x9d sparkles are visually distinguished from those caused by gas inclusions; non-gas derived sparkles appear as 8-pointed xe2x80x9cstarsxe2x80x9d, the points radiating in the 1,1,1 directions, and have no flat facets. This is attributable to their having formed so fast they did not have time to grow facets.
Another kind of xe2x80x9csparklesxe2x80x9d commonly found near the bottom of an ingot are platinum sparkles which are very small, about 1-5 xcexcm, formed as a metallic moss. Their location alone sets them apart from gas-induced, or typical non-gas derived sparkles. Such platinum sparkles would not be expected in ingots grown in graphite crucibles but for the fact that platinum is used in the preparation of growth stock.
Since solidification of even a relatively small ingot is very slow, the bottom of the charge in the crucible becomes solid, giving form to the ingot while the upper portion of the charge is still molten and tightly held within the upper portion of the walls of the crucible because the upper portion is not cool enough to shrink. The growing ingot, upon cooling sufficiently, contracts and becomes spaced apart from the graphite floor. When the upper portion is finally cool enough to shrink the ingot falls and rests on the floor of the graphite crucible. The timing of this xe2x80x9cingot dropxe2x80x9d is unpredictable. The result is that in the initial portion of growth of the ingot, its bottom is in contact with the graphite bottom of the crucible affording excellent and predictable heat transfer; in the intermediate portion of the growth, the bottom of the ingot is spaced apart from the graphite floor of the crucible resulting in poor but predictable heat transfer; and in the last portion of the growth the bottom of the ingot is again in contact with the floor of the graphite crucible affording excellent heat transfer. The unpredictability of when the heat flow changed results in poor control of growth and inclusion of gas.
As indicated above, transversely spaced apart indentation means adapted to support an ingot near its edges and maintain its bottom in spaced apart relationship with the floor of the crucible, is an effective means to control the temperature gradient through the solidifying ingot and such control is found to minimize sparkles. Another means for obtaining the controllable temperature gradient is to configure a graphite crucible so that an ingot grown from a shrinking melt is continuously in contact with the floor of the crucible, instead of spaced apart from it. Such configuration provides an equally effective solution to the problem of sparkles due to gas inclusions in an ingot. In either case, controlled solidification is maintained slowly enough to allow gas molecules to difffuse out of the lattice being formed, and upwards out of the melt.
Any crucible used to grow a macrocrystal in a furnace is specially configured for the particular purpose at hand, taking into account myriad considerations relating to the properties of the melt, the solid formed from it, the requirements of the material from which the crucible is to be formed, the size of the crucible, etc., so that the configuration of the crucible is understood to be uniquely adapted for the crucible""s intended purpose.
A configuration of a crucible to form a plate-shaped macrocrystal is disclosed in U.S. Pat. No. 5,116,456 to Nestor. The crucible is combined with a hopper integrally joined to the rim of the crucible by an inclined transition zone. Because the volume of a charge of crystalline powder growth-stock required to form a plate-shaped macrocrystal is greater than that of the macrocrystal to be grown, the hopper has a larger cross-sectional dimension than the lower crystal growing region of the crucible thus maxim the charge required to form the plate-shaped macrocrystal. The height of the furnace required to melt the charge is minimized though the hopper holds the excess volume of growth-stock, larger than the volume of the crucible, which larger volume when melted may fill the crucible to about the bottom of the hopper; and, if desired, the melt may initially extend part way into the hopper since the top surface of the melt will lower during growth of the crystal. Because of the shrinkage of volume upon melting, the height of the walls of the hopper would typically be less than about one-third (33.3%) of the height of the walls of the crucible, as shown in the ""456 reference. The transition zone is steeply inclined to allow the shrinking melt to slide into the crucible, rather than hold the macrocrystal in the transition zone, to address a problem which does not arise in the invention claimed herein.
Another crucible for growing a Group III-V semiconductor as a small macrocrystal is disclosed by Clemans in U.S. Pat. No. 4,946,542 to solve a problem unrelated to that solved by the applicant. The problem is that a Group V element (e.g. phosphorus, P) is more volatile than a Group m element (e.g. indium, In) and P tends to escape during crystal growth. Therefore a less volatile compound such as boric oxide is used as a sealing material to seal the surface of a melt (e.g. indium phosphide, InP) while it is being solidified in a small-diameter crucible, typically about 5 cm in diameter. The solid Group III-V macrocrystal is formed in combination with the sealing material supported on the top surface of the semi-conductor macrocrystal. The crucible is made to have a step extending around the periphery of its inner surface at a location near the top of the frozen crystal. The crucible has a larger cross-sectional area above the the step than below it, so that when the melt freezes, the solid has a bulge near its top which, by gravity, nests against the step to seal the lower portion of the solid crystal against entry of the liquid encapsulant. In a small crucible, smaller than 7.5 cm in diameter, the friction forces between the contiguous surfaces of melt and crucible are so strong while the melt is freezing, and before the melt is solidified, the small crystal cannot drop into the gap forming at the bottom of the crucible until it is fully solidified. Therefore, small crystals do not show xe2x80x9cflawsxe2x80x9d caused by dropping before the melt is fully solidified. In a crucible larger than 7.5 cm in diameter, the solidifying ingot drops to the floor of the crucible, or to the stepped circumferential shoulder of a conical bottom, before the ingot is solidified, the larger the crucible, the worse the effects of the drop.
It has been discovered that when a macrocrystal is grown from a shrinking melt in a crucible having an inner surface which is not wetted by the melt, solidified melt shrinks away from and separates from the sides and bottom of the crucible. After initial solidification causes this separation, it forms a barrier against flow of heat out of the bottom of the crucible while the rest of the crystal is being formed. The separation reduces the temperature gradient in the crystal already grown; a steep temperature gradient causes strain or lattice slippage which affects the optical quality of the solidified crystal. More uniformly controlled, and therefore better, cooling is obtained if the macrocrystal is grown in a smooth-walled cylindrical crucible with non-wetted walls, and the solidified macrocrystal comes to rest on the bottom of the crucible; but by then the damage is done because the gradient could not have been controlled. A gap between the bottom of ingot and the crucible is not inherently harmful unless the gap changes rapidly and affects the rate of growth. More uniformly controlled, and therefore better, cooling is not obtained if the crystal is grown in a crucible with a cylindrical upper portion and a tapered lower portion, tapered downward, and with non-wetted walls, because the solidified macrocrystal slides towards the bottom until it is supported by the stepped circumferential shoulder of the crucible""s conical bottom, leaving a gap between the lower surface of the crystal""s conical bottom, and the corresponding conical surface of the crucible""s bottom. By xe2x80x9ctapered downwardxe2x80x9d is meant that the upper portion of the crucible is wider than the lower portion. Again, the damage is done because in either case, the ability to cool the solidified crystal and reliably maintain a controllable temperature gradient between the regions at the boundary of the melt, is jeopardized.
No portion of, or horizontal plane in, a conventionally grown Stockbarger ingot of sufficient diameter and mass to drop to the shoulder of a conical-bottomed crucible, remains at all times in the same horizontal plane with respect to the crucible in which the ingot was grown. Sudden change of the relative positions of a horizontal plane through the ingot and the crucible adversely affects growth of the ingot, which growth is still not complete.
However, transversely spaced apart indentations (referred to as xe2x80x9cindentation meansxe2x80x9d) including grooves or serrations in the walls of the novel graphite crucible provide effective heat transfer control because the bottom of a solidified ingot remains spaced apart from the interior floor of the crucible, avoiding the sudden shift of weight caused by a falling ingot. These indentation means in the walls are located beneath a horizontal plane through the upper portion of the crucible, at about two-thirds (66.6%) of the vertical height of its walls, measured from the floor of the crucible (xe2x80x9cthe two-thirds planexe2x80x9d for brevity). Most preferably the indentations are located beneath a plane through one-half (50%) the vertical height of the crucible""s walls. Additional steps located above the two-thirds plane provide no noticeable improvement since any steps below the plane will support the partially solidified ingot so that it will not drop. In either case the bottom of the macrocrystal ingot is spaced apart from the bottom surface of the crucible. Effective control of heat transfer to provide the appropriate temperature gradient minimizes the formation of sparkles in an ingot grown so as to have its bottom spaced-apart from the crucible""s floor. Moreover, maintaining a gap under the ingot reduces the increase in power needed to provide directional cooling, and results in reduced strain within the crystal because there is less of a temperature gradient within the solid crystal. It follows that a novel finished ingot will be directionally solidified as it grows in a vertical direction and have protruberances formed by the reverse impression of indentations in the walls, which protruberances hold a phantom fixed horizontal plane within the ingot in a stationary position relative to a plane through the crucible; and ridges are formed on the bottom of the ingot as reverse impressions of grooves in the bottom, which ridges aid or direct the spread of a component. The protruberances extend substantially laterally for at least 50 xcexcm, and preferably range from about 0.8 mm to about 4.7 mm (height from the surface of the crystal), and are located lower than about the two-thirds (66.6%) plane, preferably lower than one-half (50%) of the vertical height of the ingot, measured from its bottom surface. The ridges extend substantially vertically downward for at least 50 xcexcm, and preferably range from about 0.8 mm to about 4.7 mm (projection from the bottom, generally lateral, surface of the crystal).
It has been discovered that grooves or serrations in the floor of a graphite crucible in which an ingot is grown from a melt in a Stockbarger controlled-atmosphere furnace, the grooves preferably being directed in the same directions favored for growth by the crystal lattice to be grown, produces a minimal number of multiplicities, if not an ingot which is an essentially single crystal. Such grooves in the floor of a conventional graphite crucible are effective to produce an essentially single crystal ingot. The object is to cover the bottom of the ingot with a grid of dendritic needles growth of which is oriented by a seed, or preferably starts spontaneously with a selection process for the fastest propagation which excludes the growth or overrides the growth of multiplicities. Grooves at least 50 xcexcm deep are preferred, more preferably at least 0.8 mm (0.03125 in) and up to about 4.7 mm (0.1875 in) deep. Scratches less than 50 xcexcm deep, or rubbing the bottom surface of a crucible can align epitaxial growth of crystals but grooves allow the melt at the bottom to be cooler than with scratches, and more susceptible to dendritic growth of any orientation.
A novel macrocrystal ingot grown from a crystallizable inorganic halide melt in a carbon or graphite crucible under earth""s gravity is essentially free from gas-induced sparkles in the upper half of the grown ingot disregarding flaws near the top, that is in the topmost portion while the ingot was growing. Typically in an ingot having an effective diameter of at least 15 cm, the uppermost 5 mm is disregarded, this being the growth during the last five hours; in addition, while the macrocrystal is being cooled, typically at a rate of 5xc2x0 C./hr (for the 15 cm diam. ingot) approximately another 5 mm below the uppermost five hours of growth is unusable because of effects of sublimation while the macrocrystal is coolingxe2x80x94the larger the diameter, the slower the cooling and the greater the effect of sublimation. Minimizing the number of multiplicities produces less than half the number of multiplicities compared to the number present in an ingot of the same salt grown from an equivalent mass of crystals of the same size in a conventional crucible of the same size and shape, except with a smooth floor. For evident reasons, it is preferred to combine the grooves in the floor of the crucible with a configuration which will afford better control of the temperature gradient than is afforded by a conventional crucible, and thus also minimize the formation of sparkles.
It is therefore a general object of this invention to provide a crucible with a grooved or serrated bottom surface, the grooves or serrations either (i) running parallel to each other in only one direction, or (ii) running parallel to each other in either two or three directions at an angle, one parallel set being typically at either 60xc2x0, 72xc2x0 or 90xc2x0 to the other parallel set, the angle chosen depending upon the desired direction of crystal growth propagation; preferably growth is in the 100 direction, but if seeded, may be in the 111 direction; though the side elevational profile of a groove is not narrowly critical, the sides of each groove are preferably angulated rather than vertical, to facilitate separation of the crystallized bottom surface of a crystal.
It is another general object of the invention to provide a crucible in which the walls support the weight of a directionally solidified macrocrystal, irrespective of the geometrical shape of the interior of the crucible; because it is critical that the melt fails substantially to wet the walls of the crucible, the weight of the growing crystal is supported by the walls of the crucible, the separation between the lower surface of the macrocrystal and the bottom of the crucible is maintained, and loss of heat through the bottom of the crucible is controlled. By securing the crystal in the novel crucible with indentation means located as specified above, so that a horizontal plane through the ingot cannot move relative to the corresponding plane through the crucible, upon solidification, all movement related to contraction of the melt will be as slow as the growth process, and under the control of the operator, by adjusting power and furnace temperature.
It is a specific object of this invention to provide either a unitary (or monolithic), or a disassemblable crucible made of a material inert with respect to a melt to be solidified, the material having a melting point substantially greater than that of the melt; the crucible is preferably of graphite or carbon, or a substitute for either, able to withstand temperatures required to melt the material to be grown; such a crucible is reusable. The crucible""s walls are essentially non-wetted by the melt, and have indentation means, located as specified above, to support a grown ingot before it is fully solidified; the configuration of the indentations is such as to provide sufficient support for the weight of solidified melt in a region below the indentations. An indentation means, located in the walls of the crucible as specified above, may be a generally lateral projection (referred to as a xe2x80x9cstepxe2x80x9d) from the interior surface of the crucible, or it may be a channel-like recess (referred to as a xe2x80x9cchannelxe2x80x9d) formed in the interior surface. The lateral width of a step or channel is empirically selected to be in the range from about 0.1% to about 1.5% of the diameter of the crystal, preferably about 1%. When the macrocrystal grown is not circular in cross-section at its widest dimension, the term xe2x80x9cdiameterxe2x80x9d refers to the widest dimension of the macrocrystal. Graphite and carbon are practical materials for a crucible and may be coated with pyrolitically deposited carbon to decrease pore size.
Because it is not possible to see what is occurring while the crystal is growing, a crucible having indentations, located as specified above, in its inner walls to support the weight of the growing crystal provides the advantage of being able to fix a level in the growing ingot relative to a plane through that level in the crucible, and being able to predict the power requirement as a function of time at each point during the growth of an ingot; and to repeat a cycle for reliably growing desirable ingots.
It is a specific object of this invention to provide a crucible in which the bottom is provided with patterns of twin sets of grooves in each semicircle of the bottom, the pattern in one semicircle being spaced apart from the other by a small distance in the range from 1 mm to about 1 cm, so as to grow side-by-side single crystals in the ingot.
It has also been discovered that xe2x80x9csparklesxe2x80x9d caused by gas inclusions may be unexpectedly minimized if not essentially negated by growing an ingot of an alkali metal halide or an alkaline earth metal halide in a graphite crucible configured to maintain the bottom of the growing ingot in intimate contact with the graphite bottom of the crucible until the ingot is fully formed; the floor of the crucible is provided with one or more grooves or serrations; the walls are relatively smooth and substantially free of indentations deep enough to support the weight of a cooling ingot; such walls near the bottom of the crucible, whether polygonal or arcuate, typically either rectangular or cylindrical, may be (i) geometrically uniform; or (ii) xe2x80x9ctaperedxe2x80x9d (or xe2x80x9ctapered downwardsxe2x80x9d); or (ii) or xe2x80x9creverse taperedxe2x80x9d (or xe2x80x9ctapered upwardxe2x80x9d); the shape of crucible chosen depending upon the choice of temperature gradient one seeks to control in the crucible. By xe2x80x9cgeometrically uniformxe2x80x9d is meant that the crucible has the same cross-section, and the walls have the same geometrical configuration, from top to bottom. The bottom of a typical Stockbarger crucible is conical and referred to in the art as being xe2x80x9ctaperedxe2x80x9d, meaning xe2x80x9ctapered downwardsxe2x80x9d, the cross-sectional area decreasing downward. By xe2x80x9ctapered upwardxe2x80x9d or xe2x80x9creverse taperedxe2x80x9d is meant that the cross-section increases downward, the walls being tapered upwards, the cross-sectional of the graphite bottom surface being greater than that immediately above the bottom surface; the effect is to control growth and eliminate erratic heat transfer. Configuring a graphite crucible so that an ingot grown from a shrinking melt is continuously in contact with the bottom of the crucible provides an effective solution to the problem of sparkles due to gas inclusions in an ingot.